Regions - Scienze Politiche Roma 3

THE DRIVERS OF INTERREGIONAL
POLICY CHOICES: EVIDENCE FROM
ITALY
Fabio Padovano
DIPES, Università Roma Tre, Roma, Italy
and CREM-CNRS, Université de Rennes1,
Rennes, France.
Introduction - 1

New theoretical developments in literature on the
determinants of transfers from CG to LG






Bailing out expectations (Rodden, 2005; Bordignon &
Turati, 2009; Josselin, Padovano and Rocaboy, 2009)
Alignment effects (Dasgupta et al., 2002)
‘Too big to fail’ effects (Wildasin, 1997)
Asymmetries in representation of local interests in
national legislatures (Porto and Sanguinetti, 2001)
Common pool situations (Persson and Tabellini, 2001)
Soft budget constraints (Quian and Roland, 1998;
Goodspeed, 2002)
Introduction - 2

…to be added to ‘traditional political
determinants’




Political budget cycles
Local political capital (Grossman, 1994)
Interest groups
…and normative welfare economics theories



provision of differentiated public goods to
heterogeneous populations
Common standards in basic services
Race to the bottom
2 problems
1.
2.



Degrees of freedom
Institutional detail
Solution to the first  panel times series-cross
section but:
If cross section composed by a variety of countries
 loose on institutional precision
 Trade off between problem 1 and 2
How NOT to do it



Avoiding institutional complexity creates
inconsistencies between results (Feld and
Schaltegger, CesIFO 2007 vs. Feld and Schaltegger,
PC 2005 vs. Feld and Kirchgassner, RSUE 2001)
Leviathan hypothesis literature another example
(Oates, 1985; Rodden, 2003; Asworth, Galli and
Padovano, 2008, 2009)
Capturing institutional changes through dummies
does not produce very satisfactory empirical
specifications (Bordignon and Turati, 2009)
Possible solutions



Development of theoretical constructs that
identify relevant institutional details (e.g.
political economy of debt creation)  more
parsimonious specifications
More cross country institutional data in this
domain (like DPI)
Selection of within country panels


Minimize institutional variance
Large data set
Merits of Italian data set






20 regions  15 RSO and 5 RSS only institutional difference
(stable)
1996-2006 no major changes in transfer policy
 No changes of expectations
 No further institutional change dummy
210 observations  data set large enough to account for
theoretical innovations
Transfer policy historically important
Bipartisan policy of progressive substitution of transfers with
‘own resources’
Health care main responsibility of regions (50% of regional
expenditures, 70% net of administration)  but there is also
more  worth looking at overall transfers
Income per family, Italian Regions
1995-2000 95% confidence
intervals, Source: BdI
Age structure by Regions - 1
Regions
Piedmont
Population
density
(n/km2)
Population by age
0-15 (%)
>65 (%)
168
12,4
22,4
37
13,2
20,2
388
13,6
19,4
71
16,1
17,7
Veneto
253
13,9
19,2
Friuli Venezia Giulia
153
12
22,6
Liguria
291
11,1
26,5
Emilia Romagna
184
12,5
22,7
Tuscany
155
12,1
23,2
Umbria
100
12,5
23,3
Italy
192
14,1
19,7
Valle d'Aosta
Lombardy
Trentino Alto Adige
Age structure by regions - 2
Regions
Population
density (n/km2)
Population by age
0-15 (%)
> 65 (%)
Marche
155
13,1
22,6
Lazio
303
13,9
19,1
Abruzzo
119
13,4
21,3
Molise
72
13,4
22
Campania
424
17,5
15,3
Puglia
209
15,7
17,3
Basilicata
60
14,5
19,9
Calabria
133
15,3
18,3
Sicily
195
16,2
18
68
12,9
17,6
192
14,1
19,7
Sardinia
Italy
Transfers in Italian public sector
(% of total expenditures, 2001)
Taxes
Soc. Sec.
contributions
Transfers from
(1)
Central government (1)
(2)
(3)
(4)
Other
revenues
(5)
Deficit
(6)
78,3
0,2
0,0
0,5
0,0
0,0
0,0
0,1
10,7
10,2
0,0
70,1
27,4
0,0
0,0
0,0
0,0
0,4
2,0
0,0
40,9
0,0
53,0
0,0
0,0
0,0
0,2
0,3
4,9
0,8
0,0
0,0
0,0
0,0
90,2
0,0
0,2
0,3
4,9
0,8
28,5
0,0
21,9
0,0
13,2
0,0
0,0
1,3
33,5
1,6
Other public institutions (6)
3,6
0,2
52,0
4,7
12,6
0,0
3,4
5,1
18,6
-0,2
Duplications
0,0
0,0
57,7
1,2
33,5
0,0
0,6
1,6
5,5
-0,1
Social security institutions (2)
Regions (3)
Local Health Units (4)
Provinces and municipalities
(5)
Transfers vs. own resources
Type of transfers, 1996-2005
Figure 3. Total, current and capital transfers
120000
Millions euro
100000
80000
60000
40000
20000
0
1996
1997
1998
1999
2000
2001
2002
ye a r
TR_ITA
TCC_ITA
TCK_ITA
2003
2004
2005
Model specification

Specification
TRit / POPit  f (POL it , ECO it , HEALTH it , DEM it )


Transfers per capita
 Total
 Earmarked to current expenditures
 Earmarked to capital expenditures
As a function of
 Economic state variables
 Political variables
 Demographic indicators
 Health care variables
Economic state variables





+ Ut-1 lagged unemployment
- DGDP/POP regional growth differential
- GDP/POP income per capita
+ TRN linear trend, incremental rule
(spesa storica)
Padovano (2007), Perotti (2001) vs.
closing income gap
Political variables - 1





+ ELN national elections dummy (Grossman,
1994; Rogoff, 1990)
+ ELR regional elections dummy (Grossman,
1994; Rogoff, 1990)
- NDIF vote margin in national elections
+ RDIF vote margin at regional elections (Cox
and McCubbins, 1988)
+ RDIF, - RDIF2 (Dixit and Londregan, 1994)
Political variables - 2



+ SAME dummy for alignment effect
(Dasgupta et al. 2002)
+ YEARS lobbying efficiency of region
(Olson, 1982)
- RIGHT dummy for ideology (Hibbs,
1977, Alesina, 1997)
Demographic variables



POP
+ demand effect, too big to fail
effect
– economies of scale
+ POP15 education, social security
+ POP65 health care, social security
Health care variables

BED, number of hospital beds × 1000
inhabitants




+ demand induced effect, Niskanen effect
- economies of scale (Crivelli et al. 2000)
+ PUPHY Public sector doctors, demand
induced effect, Niskanen effect
+ PRPHY private sector doctors,
demand effect
Empirical strategy








Model 1: only economic state variables  welfare
economics explanatory power
Model 2: full model, 20 regions
Model 3: 15 RSOs
Model 4: 5 RSSs
Model 5: current transfers
Model 6: capital transfers
Model 7: 1999-2006 sample, check for expectations
changes
Model 8: political economy vs. welfare economics
interpretation of economic state variables
Estimates: economic variables
Model
1
2
3
4
5
6
7
8
Sample
20
regions
1996-2006
20
regions
1996-2006
15
RSOs
1996-2006
5
RSSs
1996-2006
20
regions
1996-2006
20
regions
1996-2006
20
regions
1999-2006
20
regions
1996-2006
TR/POP
TR/POP
TR/POP
TR/POP
TRCC/POP
TRCK/POP
TR/POP
TR/POP
Dependent
variable
Ut-1
3.121***
(0.303)
4.4644***
(0.38)
3.446***
(0.411)
2.845***
(0.592)
0.539***
(0.072)
4.122***
(0.48)
3.928***
(0.608)
DGGDP/POP
-1.22***
(0.314)
-2.516***
(0.265)
-2.69***
(0.472)
-1.645
(0.532)
-0.895***
(0.089)
-2.407***
(0.307)
-2.636***
(0.388)
GDP/POP
-13.088
(8.259)
0.017**
(0.008)
TREND
C
0.276***
(0.023)
-2.349***
(0.326)
0.038
(0.04)
-0.086***
(0.024)
-0.161
(0.74)
0.006***
(0.002)
-0.016**
(0.007)
0.02***
(0.008)
-0.377***
(0.07)
-2.263***
(0426)
-1.865***
(0.408)
Estimates: political variables-1
Model
1
2
3
Sample
20
regions
19962006
20
regions
19962006
15
RSOs
19962006
Dependent
variable
TR/POP
TR/POP
TR/POP
4
5
RSSs
19962006
TR/POP
5
6
7
8
20
regions
1996-2006
20
regions
1996-2006
20
regions
19992006
20
regions
1996-2006
TRCC/PO
P
TRCK/POP
TR/POP
TR/POP
ELN
0.093***
(0.02)
0.103***
(0.025)
0.082***
(0.024)
0.032***
(0.004)
0.069***
(0.015)
0.103***
(0.021)
ELR
0.128***
(0.02)
-0.119
(0.207)
0.071
(0.054)
0.02***
(0.007)
0.173***
(0.017)
0.128***
(0.022)
YEARS
0.037***
(0.005)
-0.027***
(0.041)
0.025**
(0.012)
0.001
(0.001)
0.043***
(0.004)
0.039***
(0.006)
SAME
0.035**
(0.017)
0.066***
(0.014)
0.058**
(0.027)
0.006***
(0.002)
0.023
(0.014)
0.032**
(0.017)
0.05***
(0.01)
Estimates: political variables-2
Model
1
2
3
Sample
20
regions
19962006
20
regions
19962006
15
RSOs
19962006
Dependent
variable
TR/POP
TR/POP
TR/POP
NDIF
-9.806***
(2.2)
-23.9*
(14.19)
RDIF
1.126***
(0.174)
0.253***
(0.069)
(RDIF)2
-1.372***
(0.393)
RIGHT
-0.067***
(0.017)
4
5
RSSs
19962006
TR/POP
15.928***
(2.16)
5
6
7
8
20
regions
1996-2006
20
regions
1996-2006
20
regions
19992006
20
regions
1996-2006
TRCC/POP
TRCK/POP
TR/POP
TR/POP
-16.942***
(2.98)
3.16***
(0.489)
-7.971
(1.69) ***
-10.434***
(2.33)
0.234***
(0.1)
0.045***
(0.016)
0.64***
(0.155)
1.023***
(0.19)
-0.459
(0.38)
-0.896**
(0.489)
-0.066***
(0.018)
-0.066
(0.018)
-38.144***
(5.87)
-0.0005***
(0.00001)
-0.11
(0.03)
0.026***
(0.003)
Estimates: demographic
controls
Model
1
2
3
4
5
6
7
8
Sample
20
regions
19962006
20
regions
1996-2006
15
RSOs
19962006
5
RSSs
19962006
20
regions
1996-2006
20
regions
1996-2006
20
regions
19992006
20
regions
19962006
Dependent variable
TR/POP
TR/POP
TR/POP
TR/POP
TRCC/POP
TRCK/POP
TR/POP
TR/POP
-1.23-07***
(2-08)
-9.35-08***
(1.75-08)
8.89-06***
(1.59-06)
-5.49-08
(3.53-08)
-2.3-08***
(3.35-09)
-1.51-
-1.44-
07***
07***
(2.68)-08
(2.33-08)
POP
POP15
5.989***
(1.772)
0.582
(1.843)
POP65
7.178***
(0.805)
2.599**
(0.856)
28.852**
*
(11.01)
1.569
(1.157)
0.236
(0.358)
9.492***
(2.27)
5.405***
(1.756)
2.203***
(0.446)
0.707***
(0.19)
7.748***
(0.989)
6.818***
(0.84)
Estimates: health care
variables
Model
1
2
3
4
5
6
7
8
Sample
20
regions
1996-2006
20
regions
1996-2006
15
RSOs
19962006
5
RSSs
19962006
20
regions
1996-2006
20
regions
1996-2006
20
regions
19992006
20
regions
1996-2006
TR/POP
TR/POP
TR/POP
TR/POP
TRCC/POP
TRCK/POP
TR/POP
TR/POP
1.45-05
(0.7.74-06)
3.24-06***
(6.42-07)
-3.64-05***
(5.89-06)
3.41-05***
(4.89-06)
0.0003***
(3.7-05)
-0.0002
(0.0002)
0.0004**
(0.0002)
-0.161
(0.74)
-0.377***
(0.07)
-2.263***
(0426)
-1.865***
(0.408)
Dependent variable
BEDS
3.00-05***
(4.23-06)
1.68-05***
(3.5-06)
PUPHY
0.0005**
(0.0002)
0.0006***
(0.0001)
5.31-05***
1.52-05
0.276***
(0.023)
-2.349***
(0.326)
Yes
Yes
No
No
Yes
No
Yes
Yes
Adjusted R2
0.485
0.755
0.755
0.849
0.65
0.777
0.878
0.715
S.E.R.
0.374
0.408
0.451
0.16
0.23
0.104
0.409
0.404
50.92***
25.67***
30.5***
47.617***
17.96
42.75***
59.86***
23.405***
Durbin Watson
2.05
1.947
1.85
2.02
1.97
1.78
2.177
1.93
Obs.
210
210
165
55
210
210
150
210
C
AR(1)
F-statistic
RSOs fixed effects
Region
Model 3
Region
Model 3
ABR
7.26
UMB
6.41
MOL
7.06
LOM
6.38
CAL
6.98
ERO
6.31
VEN
6.71
MAR
6.29
CAM
6.67
TOS
6.26
PUG
6.66
PIE
6.25
BAS
6.49
LIG
6.06
LAZ
6.44
RSSs fixed effects
Region
Model 4
SIC
-49.86
SAR
-19.26
FVG
-16.81
TAA
-14.05
VDA
-7.19
Main results: commentary-1



Inclusion of political, health care and economic variables
increases model’s explanatory power by 33%
In RSOs electoral process prevails
 PBC
 Alignment effect
 Grants reward local political success (Cox and McCubbins,
1988)
 National political success lowers grants
In RSSs lobbying more important (different party
system)


Grants targeted to swing regions (Dixit and Londregan,
1994)
More resistance to further grants in RSSs
Main results: commentary-2

Right wing governments





receive less grants in total and for current
expenditures (partisan effect)
Receive more grants for capital expenditures
Health care variables reveal significant
induced demand/Niskanen effects
No expectations turbulence (but more
research is warranted)
Political economy explanations of
interregional redistribution more supported
than standard welfare economics ones